Bayesian filters need tuning by the individual user
And sometimes they are just not very good. I tried Mozilla's version of Bayesian filtering a few weeks ago (while I was still using Moz as my mail client; I am not any more), but in my case I simply don't get enough spam on my current primary address to make the filter trainable. (And I am unconsolable, of course.) I was getting close to 100% false-positives, and the spam was still ending up un my inbox.
My own set of filter rules catch over 99% of spam, according to these rules...
Whitelist (of course) Filter by TLD (kill all.kr,.ru etc. mail) Filter by buzzwords e.g. "this is not spam", "click here", "for more info"... Filter by header fields (known open relays or dodgy domains) and a "kill these scum" filter to deal with specific offenders, based on domain names in the body of the email.
I've been using these rules for several years, and they trap almost all spam, even with the obfuscated html that is currently trendy.
Of course, none of this is doing anything about the real problem, since the mail is still getting through to my computer. Hence, I still filter stuff to a spam folder, just to be sure I don't score any false-positives.
You have a point. In the article, Barr seems to indicate that his main problem with the earlier release is that he didn't RTFM. It does seem churlish for him to continue to berate the developers for his own laziness.
I often wonder what this is all about. On the one hand I see an argument for a distribution that has everything compiled from source (of which I would thoroughly approve if I didn't have to maintain computers of more than one architecture). And on the other hand, I see numerous posts from individuals who don't want to get their hands dirty and do their own compiling.
This trite emerge stinkyfinger that we see trotted out in this type of discussion is no more enlightened than rpm -U stinkyfinger.
And no, I'm not a RedHat zealot (I prefer Slackware).
but for some reason mplayer isnt included in the debian distribution
It might have something to do with the fact that (as it says somewhere in the documentation) mplayer sort of has to be built individually for it to work properly.
I don't know if that's actually true, though. I wouldn't be surprised if there were licencing issues relating to those Win32 codecs that render the thing too "impure" for the Debian ethos. FWIW, it doesn't get included in Slackware either.
To be honest, I haven't really used xine much, as I had issues with it crashing a while back. But MPlayer is truly a great program which handles every type of movie file I've come across (so far).
The article's right where it says installation can be non-trivial, though. That's why I've still got 0.90 running on my computers. I'll get around to upgrading eventually, but to be honest, I think it's as good as it gets...
Those Opinel knives are the best in the world. They take a good edge, they don't fold up on to your fingers unexpectedly, and they look enough like a workman's knife (because that's what they are) not to inspire the cupidity of light-fingered individuals.
These Turing tests usually munge the appearance of the characters in some way.
In practice, you could just have a jpeg of your address in 42-point boldface Helvetica, which anybody can read if they're not totally blind, and this would be an adequate foil for the majority of bots.
Not all Linux distros use SysV init scripts. Slackware uses BSD-style scripts which are a lot easier to get your head around. (They still depend on runlevels, though...)
Since Microsoft's code has repeatedly (ad nauseam) been demonstrated to be as full of holes as a Swiss cheese and apparently almost infinitely susceptible to viruses crafted by the most juvenile of script-kiddies...
What value is implicit in being able to read their source code?
akin to language fascism and will ultimately lead to a rigid linguistic structure with no room for growth and change.
And I will gladly slit a language fascist's weasand from ear to ear, quoting "Thou remnant!" the while...
[cough...never mind]
but anyway... since I have lived here in Australia for 16 years, I have got used to a fairly healthy situation where British spellings and pronunciations are preferred, but Americanisms accepted where the context or word is appropriate.
This company is asking for a whole lot of attention from black-hat crackers
True, but I would be more worried about the security credentials of the company. Given that many security firms have had experience with staff of negotiable honesty - and they have to have police clearance (at least here in Australia) I would be intensely suspicious of this type of company.
For this kind of service to be useful, it would have to be hitched up to a heavy-duty encryption algorithm and have an equally heavy-duty audit trail listing everybody who has had contact with the mail.
In my own case (even though my mail is probably quite boring to others, and I don't have anything in particular to hide) I still wouldn't like the feeling that someone has read my mail first.
Perhaps this could be extended to other services; Douglas Adams once referred to the Greek ferry timetables as "the greatest work of fiction known to man", and others have drawn the same parallel to British Rail timetables...
I've often wondered about Microsoft's OSs booting so quickly. I can put up with it by simply not running Windows, and never rebooting Linux except when I build a new kernel.
But I guess if it's a choice of being able to boot in no time at all while being aware that the machine can blue-screen in about the same interval, I'll take the longer boot-time...
When you say "some people have too much time on their hands", are you talking about the guy that made the case mod, or the guy that spent 10 minutes waiting for pictures of a case mod?;)
One could argue a case for both.
My preference for a case is one that sits happily under my desk and is both well-ventilated and quiet.
Given the current doubts regarding Microsoft OSs' resistance to viruses or security threats, I personally will not be quaking in my boots at the prospect of the world switching to Micro$oft from linux. If anybody does, then more fool them.
Amanda is a happy-go-lucky ignoramus who has not got a single clue about any of the ministries we have had the misfortune of her heading up
That is probably the most charitable description of her that I would come up with.
The present Cabinet might be best characterised as a bunch of malicious headkickers, and Vanstone is no exception. She is several stages more cretinous than her cronies, though.
Slashdot Mirror
And sometimes they are just not very good. I tried Mozilla's version of Bayesian filtering a few weeks ago (while I was still using Moz as my mail client; I am not any more), but in my case I simply don't get enough spam on my current primary address to make the filter trainable. (And I am unconsolable, of course.) I was getting close to 100% false-positives, and the spam was still ending up un my inbox.
My own set of filter rules catch over 99% of spam, according to these rules...
Whitelist (of course) .kr, .ru etc. mail)
Filter by TLD (kill all
Filter by buzzwords e.g. "this is not spam", "click here", "for more info"...
Filter by header fields (known open relays or dodgy domains)
and a "kill these scum" filter to deal with specific offenders, based on domain names in the body of the email.
I've been using these rules for several years, and they trap almost all spam, even with the obfuscated html that is currently trendy.
Of course, none of this is doing anything about the real problem, since the mail is still getting through to my computer. Hence, I still filter stuff to a spam folder, just to be sure I don't score any false-positives.
I agree. We should just kill them instead. As cheaply as possible. Stoning? Drowning?
You have a point. In the article, Barr seems to indicate that his main problem with the earlier release is that he didn't RTFM. It does seem churlish for him to continue to berate the developers for his own laziness.
I often wonder what this is all about. On the one hand I see an argument for a distribution that has everything compiled from source (of which I would thoroughly approve if I didn't have to maintain computers of more than one architecture). And on the other hand, I see numerous posts from individuals who don't want to get their hands dirty and do their own compiling.
This trite emerge stinkyfinger that we see trotted out in this type of discussion is no more enlightened than rpm -U stinkyfinger.
And no, I'm not a RedHat zealot (I prefer Slackware).
It might have something to do with the fact that (as it says somewhere in the documentation) mplayer sort of has to be built individually for it to work properly.
I don't know if that's actually true, though. I wouldn't be surprised if there were licencing issues relating to those Win32 codecs that render the thing too "impure" for the Debian ethos. FWIW, it doesn't get included in Slackware either.
This is true, but remember that those of us in *NIX land don't have to put a .EXE on the end of a filename to make it a program.
"mplayer" is sufficient.
The article's right where it says installation can be non-trivial, though. That's why I've still got 0.90 running on my computers. I'll get around to upgrading eventually, but to be honest, I think it's as good as it gets...
Those Opinel knives are the best in the world. They take a good edge, they don't fold up on to your fingers unexpectedly, and they look enough like a workman's knife (because that's what they are) not to inspire the cupidity of light-fingered individuals.
In practice, you could just have a jpeg of your address in 42-point boldface Helvetica, which anybody can read if they're not totally blind, and this would be an adequate foil for the majority of bots.
Not all Linux distros use SysV init scripts. Slackware uses BSD-style scripts which are a lot easier to get your head around. (They still depend on runlevels, though...)
Thus quoth the junk-mailer. No wonder you're posting anonymously.
Do you, perhaps, mean that you have rights at the weekend but not on weekdays?
Enquiring minds want to know... :-)
Indeed it was... at least if I remember the quote correctly.
What value is implicit in being able to read their source code?
You mean it's handy to have someone you can beat the crap out of (err, someone out of whom one may beat the excrement) if he transgresses... :-)
For what it's worth, I agree.
And I will gladly slit a language fascist's weasand from ear to ear, quoting "Thou remnant!" the while...
[cough...never mind]
but anyway... since I have lived here in Australia for 16 years, I have got used to a fairly healthy situation where British spellings and pronunciations are preferred, but Americanisms accepted where the context or word is appropriate.
Sounds like you need to get a decent mailbox with a lock on it.
True, but I would be more worried about the security credentials of the company. Given that many security firms have had experience with staff of negotiable honesty - and they have to have police clearance (at least here in Australia) I would be intensely suspicious of this type of company.
For this kind of service to be useful, it would have to be hitched up to a heavy-duty encryption algorithm and have an equally heavy-duty audit trail listing everybody who has had contact with the mail.
In my own case (even though my mail is probably quite boring to others, and I don't have anything in particular to hide) I still wouldn't like the feeling that someone has read my mail first.
Perhaps this could be extended to other services; Douglas Adams once referred to the Greek ferry timetables as "the greatest work of fiction known to man", and others have drawn the same parallel to British Rail timetables...
But I guess if it's a choice of being able to boot in no time at all while being aware that the machine can blue-screen in about the same interval, I'll take the longer boot-time...
One could argue a case for both.
My preference for a case is one that sits happily under my desk and is both well-ventilated and quiet.
My kernel load is a lot quicker than the POST and BIOS loading, which takes > 45 seconds.
If you never power off your machine, you (almost) never have to reboot it. So net time saved is much greater...
Given the current doubts regarding Microsoft OSs' resistance to viruses or security threats, I personally will not be quaking in my boots at the prospect of the world switching to Micro$oft from linux. If anybody does, then more fool them.
That is probably the most charitable description of her that I would come up with.
The present Cabinet might be best characterised as a bunch of malicious headkickers, and Vanstone is no exception. She is several stages more cretinous than her cronies, though.